Compatibility: The most well-known ZK scaling projects promise EVM compatibility but run a different VM architecture that requires developers to transpile or migrate code, breaking existing developer tooling and infrastructure. This is a disadvantage, because it means that they'll need to build a new ecosystem of developer tools and infrastructure from scratch.
The Ethereum developer ecosystem is the result of a huge investment of time and energy, and preserving compatibility with it is a major selling point for the zkEVM. We're not just scaling transactions per second on Ethereum, we're also building on the foundation of a vibrant community and ecosystem.
Performance: It's difficult to provide performance comparisons as benchmarks aren't public or don't exist yet. However, we anticipate that the Polygon zkEVM prover is the fastest one, being able to validate 500K gas units on a single CPU server (64 cores) in about 5 minutes time.
Polygon was the first to implement recursive STARKs, a core building block in ZK L2s, almost a year ahead of other teams. We've seen huge ~100x improvements in the most expensive operations in the zkEVM: keccak, ECDSA, and recursion, over the previous state of the art. Enabling recursion, zkEVM has a clear path to improve ZK proof times and scalability.
Security and Ethereum-alignment vs alt-L1s: Using ZK allows a ZK L2 to inherit the security of Ethereum. ZK proofs ensure that every transaction on the zkEVM is valid. This ensures that it's impossible for operators of the L2 to steal funds from users. Security (as well as scalability) is a major benefit for the zkEVM relative to alt-L1s, and it ensures that the zkEVM is closely aligned with Ethereum, still by far the dominant chain in terms of developer activity, TVL, NFT activity, etc.